What Does an AI SEO Agency Actually Do? (And What It Cannot)
An AI SEO agency is not a shop that installs a plugin, feeds your website into ChatGPT and sends you a screenshot of a mention. That is marketing theatre, not a service. What it actually does is extend established search optimisation, technical structure, content quality, authority signals, into the newer environment of AI Overviews, chat-based assistants and answer engines, then measure whether any of it is working. For a business, the stakes are real: get this wrong and you spend a marketing budget chasing a result nobody can verify.
You are probably reading this because one of three things happened. A competitor started showing up in Google's AI Overviews or in a ChatGPT answer and you did not, and someone in the business asked why. Or you ran this through a previous agency and got a vague report full of screenshots and no method behind them. Or you are simply trying to get ahead of a shift in search before it costs you leads, the way businesses that ignored mobile-first indexing or Core Web Vitals eventually paid for it.
Each of those situations lands on the same question. What, specifically, does this work involve day to day, and where does the honest limit of an agency's influence actually sit? That is what the rest of this article answers.
What the work actually involves
The foundation has not changed much, and that is deliberate. Crawlability, clean site architecture, fast load times and clear internal linking still decide whether search engines and AI crawlers can read your site at all. On top of that sits content built to answer real questions directly, structured with schema markup so a machine can parse what the page is actually about, not just what it says.
What is new is the layer above it. AI search visibility means actively tracking whether your brand, product or service gets surfaced and cited inside AI-generated answers, across Google AI Overviews, ChatGPT, Perplexity and Microsoft Copilot. That includes technical groundwork most businesses have never heard of, such as whether a site needs an llms.txt file to help large language models understand and cite it correctly. It also includes writing content that reads as a direct, well-sourced answer rather than a page built purely to rank.
Google has been explicit that AI-generated content is judged against the same quality expectations as any other page: useful, accurate, and written for people rather than to game a system (Google, 2023). An AI SEO agency that treats this as a separate discipline, with separate rules, is building on the wrong assumption. Salt's organic search service treats traditional SEO and AI search visibility as one connected body of work, because in practice they draw on the same signals.
How visibility across AI platforms gets tracked
This is the part most agencies skip, because it is slower and less photogenic than a single screenshot. An AI answer is not a fixed result the way a page-one Google ranking used to be. Ask the same question five times, across five sessions, on two different platforms, and you will get five different answers, with different sources cited each time.
That means the right question is never "does my business appear in AI search?" It is "how often does my business appear, for which queries, and on which platforms?" Tracking this properly means running the same core topics through multiple AI tools repeatedly, logging which sources get cited and how consistently, and cross-referencing that against referral data. Google Analytics 4 can now isolate traffic arriving from AI platforms as a distinct channel, and tracking AI search traffic in GA4 is one of the more reliable ways to see whether visibility work is converting into actual visits.
A single good answer on a single day tells you almost nothing. A pattern across weeks and dozens of prompts tells you whether the work is holding. This is a frequency problem, not a yes-or-no test, and any agency reporting it as a pass or fail is simplifying it past the point of being useful.
Where the platforms genuinely differ
Google AI Overviews, ChatGPT, Perplexity and Copilot are not the same product wearing different branding. They pull from different indexes, weight authority and recency differently, and in some cases browse the live web while in others lean on a training snapshot that can be months old. Google has published guidance confirming that AI Overviews source and cite content much the way traditional search ranking does, drawing on crawled, indexed pages that meet its existing quality signals (Google, 2024).
That matters because a tactic that improves visibility in Google's AI Overviews, strong structured data, clear topical authority, will not necessarily shift how often a chatbot with a different sourcing model mentions you. Some platforms favour sites with long-established authority on a topic. Others lean toward recently published, clearly structured content regardless of domain age. How AI Overviews are changing search in Australia is worth reading in full if you want the detail on how this plays out locally.
An agency working across this space needs to test each platform on its own terms rather than assume one optimisation approach covers all of them. That is slower work than running a single audit, but it is the only version of the work that produces a defensible answer about where your visibility actually stands.
What an AI SEO agency cannot do
Here is the part most sales pitches leave out. No agency can guarantee your business will be cited inside a specific AI answer, because none of the major platforms sell, publish or expose a mechanism to buy or lock in that placement. That is a meaningful difference from paid search, where a bid and a budget buy a guaranteed position. There is currently no equivalent lever inside an AI-generated answer itself.
An AI citation is also not a ranking position you can defend once won. Models retrain, sourcing logic shifts, and a citation that appeared reliably last month can disappear with the next update, through no fault of the work done to earn it. An agency that promises permanent placement is either misunderstanding the mechanics or hoping you will not ask how it actually works. Will AI replace marketing agencies covers a related version of this same honesty gap.
What a competent agency can do is improve the underlying signals that make citation more likely, more often, across more platforms, and measure the pattern over time. That is the realistic version of the service, and it is the one worth paying for.
Scope varies a lot from one business to another. A professional services firm competing on a handful of high-value queries needs a different depth of AI visibility work than a retailer with thousands of product pages, and a conversation with Salt is usually the fastest way to work out which version applies to your situation, rather than buying a generic package built for neither.
Frequently Asked Questions
How is an AI SEO agency different from a traditional SEO agency?
The overlap is larger than most people expect. Both rely on technical SEO, structured content and authority signals, and Google has stated that AI-generated and AI-surfaced content is held to the same quality expectations as any other result (Google, 2023). The difference is the added layer of testing and tracking visibility specifically inside AI answer engines, which traditional SEO reporting was never built to measure.
How long does it take to see results in AI search visibility?
There is no fixed timeframe, because AI platforms retrain and adjust sourcing behaviour on their own schedules, not yours. Expect the same general pattern as organic SEO, gradual improvement over months rather than an overnight shift, tracked as a trend across repeated testing rather than a single result.
Can an agency guarantee my business appears in ChatGPT or Google AI Overviews?
No, and any agency that promises this should be questioned closely. There is no paid mechanism to buy a citation inside an AI-generated answer on any major platform, and sourcing behaviour can change with a single model update (Google, 2024). What can be improved is the likelihood and frequency of citation, not a guaranteed outcome.
What does an AI SEO agency actually report on?
A credible report covers crawlability and technical health, content and structured data coverage, and a tracked pattern of AI citations across platforms and prompts over time, alongside any referral traffic that AI platforms generate in GA4. A single screenshot of one chatbot answer is not a report.
Who oversees this work, a specialist or my regular marketing team?
It depends on how the agency is structured. At Salt, this sits inside the broader organic search team rather than a separate bolt-on product, because AI visibility and traditional search visibility draw on the same technical and content foundations. You can see how the wider team approaches integrated work on the About page.
AI search is not a separate channel bolted onto the side of a marketing plan. It behaves more like a new distribution layer sitting on top of the same content, structure and authority work that has always driven organic performance, which means the businesses that treat it as connected to their broader marketing, rather than as a standalone fix, are the ones most likely to show up consistently as the platforms themselves keep changing.
References
Google (2023) Google Search's guidance about AI-generated content, Google Search Central. Available at: https://developers.google.com/search/blog/2023/02/google-search-and-ai-content
Google (2024) AI Overviews and your website, Google Search Central. Available at: https://developers.google.com/search/docs/appearance/ai-overviews